Ondas reported record second-quarter revenue of $83.8 million on August 13, up 67% sequentially from $50.1 million and more than thirteen times the $6.3 million it booked a year earlier, and raised its full-year target to $525 to $550 million. The stock fell roughly 9% on the print and closed Friday at $9.24, recovering 3.7% on the day as new tariffs on imported drones lifted the entire domestic group. Market capitalisation is $5.27 billion on 570.6 million shares outstanding. The twelve-month range runs from $3.20 to $15.28, which is the honest description of what owning this has been.
The number that decides 2026 is not in the release. First-half revenue was $133.9 million. Third-quarter guidance is $140 to $155 million. Subtract both from the full-year range and the fourth quarter has to land between $236 million and $276 million, roughly $256 million at the midpoint. That is three times the record quarter just reported and close to half of the entire year compressed into thirteen weeks. It also requires converting more than half of the $757 million pro forma backlog into recognised revenue inside six months, on physical hardware: loitering munitions under the Army’s Lethal Unmanned Strike award, the ULTRA and IonStrike platforms arriving through DZYNE, and the $140 million combat engineering vehicles programme that management says begins volume delivery in the fourth quarter. Nothing in that list is a subscription. The full-year number is a shipping schedule, and shipping schedules slip.
The second thing to understand is that Ondas is not profitable, whatever the screens say. Trailing twelve-month net income shows as $85.8 million and the trailing multiple prints at roughly 48 times. Both come from a $1.04 billion warrant liability that is remeasured at fair value every reporting period. That single line is 74% of total liabilities. It generated $404.2 million of other income in the first quarter, which is how a company with $50.1 million of quarterly revenue reported $361.2 million of net income, and it is why accumulated deficit fell from $368 million at year end to $94 million by June. The company says as much in its own release, noting that the resulting swings are unrelated to operating performance or cash flow. Underneath the mark, second-quarter adjusted EBITDA was a loss of $50.6 million, and company-wide adjusted EBITDA profitability is not guided until the fourth quarter of 2027.
Dilution is the cost line that does not appear as a cost. Shares outstanding went from 380.8 million at year end to 529.8 million at June 30 and 570.6 million today, an increase of about 50% in seven and a half months. Weighted average basic shares in the quarter were 500.7 million against 150.7 million a year ago, a factor of 3.3. Set that against the 13.4-fold revenue increase and growth per share is roughly four times, not thirteen. Management’s own pro forma organic figure was 85% year over year in the quarter, and the guidance implies greater than 30% organic for both the third quarter and the year. Reported growth is accelerating while organic growth decelerates. The gap between those two numbers is the acquisition engine, and the engine runs on equity.
Stock-based compensation was $69.1 million in the quarter, $67.6 million of it in operating expenses, elevated by the vesting of executive awards. That is 82% of revenue. It is also larger than the entire adjusted EBITDA loss from which it is excluded. Total operating expenses of $199.1 million included $105.8 million of non-cash items and $4.4 million of transaction costs, which leaves adjusted cash operating expense at $93.2 million against $36.1 million of gross profit. Gross margin was 43.1% on a reported basis, down from 49.2% in the first quarter and 53.1% a year ago, with management guiding lower still in the second half on product mix and absorption across roughly 230,000 square feet of recently acquired and underutilised manufacturing space. The ramp that delivers the revenue target is the same ramp that compresses the margin on which the target is priced.
The moat is narrower and better than the story around it. The defensible asset is Sentrycs and its Cyber-over-RF approach, which takes control of an intruding aircraft by its own protocol rather than jamming the link. That distinction is not engineering vanity. Broadband jamming is effectively unusable over populated civilian airspace, which means protocol-based mitigation is close to the only lawful path to actual interdiction at a stadium, an airport or a border crossing. That is a regulatory moat, and regulatory moats last longer than technical ones. The evidence sits in the customer list: counter-drone coverage at a majority of the North American World Cup venues, the Jacksonville Jaguars becoming the first NFL franchise to go past detection to controlled mitigation, and Lockheed Martin integrating the capability into its Sanctum platform, which is the kind of relationship that survives a change in procurement leadership. Almost everything else is procurement position rather than proprietary capability. The $982 million Lethal Unmanned Strike vehicle is an indefinite-delivery award against which Ondas has captured about $240 million, which is a hunting licence with a good hit rate, not a contract. The NASA stratospheric ceiling rising from $45 million to $395 million is capacity to order, not an order. The balance sheet is offered as the third advantage, and it is the weakest, because it is circular: equity bought fourteen businesses, those businesses produce the growth rate, and the growth rate supports the multiple at which more equity is sold. Cut the share price in half and the acquisition currency halves with it.
On valuation, $5.27 billion against the midpoint of this year’s guide is 9.8 times current-year revenue. Net of the roughly $1.07 billion of cash remaining after the $325 million spent closing DZYNE and Cyberhawk, enterprise value is near $4.2 billion, or 7.8 times 2026 revenue at a 43% gross margin with negative EBITDA for another five quarters. That is a software multiple resting on a defence hardware cost structure. Anyone netting the cash should also note that a portion of it arrived attached to the warrants now carried at $1.04 billion, so the same instruments that funded the balance sheet will expand the share count against which it is measured. Consensus sits at $19.28 across nine analysts with no sell ratings, one house maintained $25 immediately after the print, and Roth initiated at Buy with $13 on August 10, five days before the quarter that would have informed it.
The base case is $8 to $11, where the stock chops between order announcements and the group’s beta of 2.75, with roughly 40% of the float short and 97 million shares changing hands on Friday alone. Price discovery here is a flow problem more than a modelling problem. The bull case retests $15.28 if the third quarter lands at the top of guidance and fourth-quarter deliveries on the Army munitions programme and the vehicle contract are visibly on schedule by the November print, at which point the short base becomes the fuel rather than the ceiling. The bear case is a cohort derating to $5 or $6 alongside AeroVironment, Kratos and the smaller drone names, triggered not by a demand shock but by a single quarter of delivery slippage: push $60 million of fourth-quarter shipments into the first quarter of 2027 and the full-year number breaks, organic growth prints in the twenties, and there is no earnings floor underneath because there are no earnings.
The disclosure that matters is the warrant footnote in the second-quarter 10-Q. Count, strike and expiry set the true diluted denominator, and every multiple in this note divides by it.